Assessment of spectral matching approaches considering near-fault ground motions in nonlinear time history analysis

Abstract

This thesis investigates the applicability of spectrum matching approaches for near-fault ground motions in nonlinear time-history analysis. Near-fault earthquake records exhibit distinctive characteristics such as strong directivity effects, pronounced velocity pulses, and highly directional energy content. These features may govern structural response and lead to seismic demands that differ significantly from those associated with far-field ground motions. As nonlinear time-history analysis is increasingly adopted in performance-based seismic design, the selection and modification of ground motion records capable of realistically representing near-fault effects has become a critical issue in engineering practice. Spectrum matching is widely used to improve the compatibility of selected ground motion records with code-based or site-specific target spectra and to reduce record-to-record variability in nonlinear analyses. Despite its practical advantages, the application of spectrum matching to near-fault ground motions remains controversial, as the frequency-dependent modifications required to enforce spectral compatibility may alter physically meaningful wave characteristics. In particular, changes in velocity pulse shape, energy concentration, and directional properties may affect the reliability of nonlinear time-history analysis results when near-fault effects dominate structural response. Within this context, this study focuses on evaluating whether commonly used spectrum matching approaches can achieve spectral compatibility while maintaining the essential characteristics of near-fault ground motions. To address this objective, a selected set of near-fault ground motion records is analyzed using both uni-directional and bi-directional spectrum matching approaches. The analyses are conducted within a nonlinear time-history analysis framework using single-degree-of-freedom systems, which allow fundamental response characteristics to be examined in a controlled and interpretable manner. The effects of spectrum matching are evaluated through response spectrum comparisons, time-history analyses of acceleration and velocity records, wavelet-based time–frequency assessment, and the examination of nonlinear response quantities such as deformation and ductility demand. This multi-level evaluation enables the influence of spectrum matching to be assessed not only in terms of spectral agreement, but also with respect to time-domain behavior and resulting structural response. The results demonstrate that spectrum matching is effective in improving agreement with target response spectra; however, this improvement may be accompanied by notable modifications in near-fault ground motion characteristics. The analyses show that strict enforcement of spectral compatibility can lead to partial distortion of velocity pulse features and redistribution of energy content, particularly within specific period ranges that are critical for near-fault response. These effects highlight the limitations of evaluating spectrum matching performance solely based on spectral criteria, especially for pulse-dominated ground motions. A comparative assessment of uni-directional and bi-directional spectrum matching approaches indicates that their performance differs depending on the characteristics of the original records. Bi-directional spectrum matching generally provides improved spectral agreement across both horizontal components and leads to more balanced modifications in the time domain. In many cases, this coordinated treatment results in smoother time-history behavior and reduced artificial oscillations when compared to the uni-directional approach. However, the results also indicate that bi-directional spectrum matching does not fully eliminate distortions in records dominated by strong velocity pulses. For such cases, both matching approaches may compromise certain near-fault features when strict spectral compatibility is enforced, suggesting an inherent trade-off between spectral agreement and physical realism. The findings further demonstrate that reliance on spectral compatibility alone may be insufficient for reliable nonlinear time-history analysis under near-fault conditions. While spectrum matching can improve consistency with target spectra, additional evaluation of time-domain characteristics and near-fault-specific features is necessary to ensure physically meaningful representation of seismic demand. In this regard, the study emphasizes the importance of considering velocity pulse integrity, energy localization, and directional effects alongside conventional spectral measures. Finally, the analytical results are interpreted in relation to existing seismic design code practices, particularly those adopted in ASCE standards and TBDY-2018. The findings suggest that while spectrum matching should not be categorically excluded for near-fault applications, its use requires careful interpretation and, where appropriate, additional guidance to balance spectral compatibility with the preservation of near-fault ground motion characteristics. Overall, this study contributes to a clearer understanding of the conditions under which spectrum matching approaches may be applied more reliably in nonlinear time-history analysis involving near-fault ground motions.